Design And Implementation Of Police Address Matching System In Shanghai

Addresses play a vital role in daily life as well as government management and are one of the fundamental components in the development of smart cities.An important piece of information that police officers need to know when handling civil reports is the accurate address of the incident,so that they can dispatch police forces to the scene as soon as possible.However,the addresses provided by the public are often not standard,making it difficult for the dispatcher to match them quickly and accurately to the corresponding standard address.Therefore,developing an address system that can match various types of addresses described by the reporting person to standard addresses is of great significance in improving the efficiency of police work.The difficulty in address matching lies in address recognition and the need for a comprehensive standard address database.Chinese addresses do not have clear word boundaries,and the hierarchical relationship between address elements may not be clear,for example,there are cases where the road name contains the city name.In addition,Shanghai has a large land area and rapid urban development,resulting in a large amount of address data that is updated quickly,further increasing the difficulty of address recognition.The existing standard address database in Shanghai has outdated data,errors,and a lack of administrative division information.Additionally,a large amount of point of interest information commonly used by the public cannot be found in the database,limiting the accuracy and reliability of address matching.To address these issues,this paper presents a BERT-based solution for address recognition.Furthermore,it enhances the standard address database in Shanghai and devises an effective address matching algorithm.In addition,this paper develops a Shanghai police address matching system.The main contributions of this study are as follows:（1）A BERT-based address recognition model is designed,with innovative tag revision and administrative division revision modules.The BERT model can identify address elements effectively by obtaining the contextual connections and address features.The tag revision module considers the hierarchical nature of addresses and address synonyms,which improves the effectiveness of address recognition and leads to an F₁ score of 0.9402.The administrative division revision module can preliminarily verify and correct the administrative division information in the original address to assist in standardizing addresses.（2）The standard address database in Shanghai is improved by redesigning the standard address table and adding a point of interest table.The original standard address table is redesigned with added administrative division information and corrected data errors.In addition,the point of interest data is obtained using map software interfaces to construct a point of interest table.（3）A multi-strategy address matching algorithm is designed,combining the point of interest table and standard address table.Different address matching methods are used depending on the types of address elements included,queries are performed using the standard address database,and results are filtered using string and spatial similarity.（4）A Shanghai police address matching system is designed and implemented,with features such as address standardization,map search,database management,and user management,which can effectively assist in case handling and improve work efficiency.